Journal of Applied Research and Technology

ISSN: 1665-6423
jart@aleph.cinstrum.unam.mx
Centro de Ciencias Aplicadas y Desarrollo
Tecnológico
México

Velázquez Arcos, Juan Manuel; Ramírez Rojas, Alejandro

Time reversibility in acoustic signals
Journal of Applied Research and Technology, vol. 2, núm. 2, august, 2004, pp. 107-115
Centro de Ciencias Aplicadas y Desarrollo Tecnológico
Distrito Federal, México

Available in: http://www.redalyc.org/articulo.oa?id=47420202

How to cite
Complete issue
More information about this article
Journal's homepage in redalyc.org
Scientific Information System
Network of Scientific Journals from Latin America, the Caribbean, Spain and Portugal
Non-profit academic project, developed under the open access initiative
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

TIME REVERSIBILITY IN ACOUSTIC SIGNALS

J.M. Velázquez , A. Ramírez & C. A. Vargas

Area de Física de Procesos Irreversibles,

Departamento de Ciencias Básicas, UAM-A,
Av. San Pablo 180, Del. Azcapotzalco, 02200, D.F.,
México. E-mail: jmva@correo.azc.uam.mx

th rd
Received: February 26 2002 and accepted March 3 2003

ABSTRACT

In this work we make a description of the time reversion problem in sound waves. Our objective is to explain the
phenomena through the Fourier transform of the Green’s function. With this function it is possible to characterize
the propagation of the emitted signal. It also can be used to express the time-reversed signal in such a way that we
can select precisely the destination site of the signal. Finally, we show some possible practical applications of this
problem.

KEYWORDS: Green’s Functions, Time Reversal Process, Inverse Scattering Analysis, Acoustics.

1. INTRODUCTION

Time reversibility arises from the second order time derivatives of the wave equation and is different from the
phenomena represented by the nonlinear behavior of media, known as inverse acoustic or electromagnetic scattering,
studied in other sites [1]. This subject gives rise to an interesting variety of items. One of them is the time reversal of a
signal by a sound mirror (TRM). Another may be the possibility to send a message with a precise physical destination.
Very interesting and useful applications might be based on the time reversibility of sound waves. In medicine, it is
possible to build devices that can destroy a brain tumor by focusing ultrasound waves; undersea communications can
be improved; in material and hydrodynamics analysis we can detect small changes in the isotropic properties of the
studied media.

The importance of time reversed acoustics is a great motivation for developing a practical generalized treatment of
this phenomena, that can be systematically applied to a broad class of scenarios. In such a treatment we obtain
expressions which take into account the geometry and nature of the original signal, and that can be used directly to
measure some data needed to reverse the time in acoustic signals. The principal features of this work are firstly to
obtain an equation, in the discrete case, that relates the Fourier transformed initial signals at the precise emitting sites
to the measured Fourier transformed signals at the precise receiving sites through the discrete Fourier transform of
the complete Green’s function. Then, by using this relation, to describe the detailed procedure involved in both the
time reversion and the sending of messages toward specific places without dispersion. The procedure resembles the
well-known technique in quantum mechanics collision theory that relates the incoming wave function with the
scattered wave by means of the complete Green’s function of the problem. In fact, we develop the Neumann series
of our discrete problem and calculate the sum in order to obtain our results.

107
Journal of Applied Research and Technology
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

Once we have sketched the formalism involved, we then comment on some possible applications.

Figure 1. Locus of transducers

2. DESCRIPTION OF THE SYSTEM

We make the assumption that our system is confined to a region of space which contains a set of devices located at
{r s } (s stands for site at the interior). These represent a set of either transmitters or receptors of a sound signal, at
our convenience. At the boundary of this region is located another set of devices, at points {r b } (b stands for
boundary), whose operation is identical to the former set, that is, a set of transducers. The transmitters at {r s } may
be considered themselves as scattering points for the signal generated by other transmitters. If the signal starts at the
inner r s as u s (t ' ) , we can record the corresponding signal at location r b as ub (t ) and, by time reversal, recover
us (t ' ) . On the other hand, we could have a problem with a different purpose in which we can emit a signal ub (t ' )
from a point r b at the boundary and send the message only to the inner site r s as u s (t ) . The sound signals are
assumed to satisfy the wave equation

∂ 2u ( r , t )
k (r ) = ∇ 2 ( u ( r , t ) / ρ ( r )) (1)
∂t 2

Here ρ (r ) is the density and k (r ) is the compressibility of the transmitting medium. This equation is time reversal
invariant since the time derivative is of second order. Then, for every u ( r , t ) there exists a signal u ( r ,−t ) that retraces
the complex paths and converges in synchrony at the original source. Suppose that a pulse u ( r , t ) leaves a source at
r s ; then we can perform physically the time-reversal process by recording by means of the transducers {r b } the
signal that arrives at the boundary during a time Τ and thereafter sending the time-reversed signal u ( r , Τ − t ) . Each
of the {r b } emit the signal ub (Τ − t ) also during a time Τ, so that the resultant field is the original u ( r , t ) focused
at r s .

One may want to make a sort of inverse application to the procedure described above: first we send a message
ub (Τ − t ) from the boundary at r b and make a similar translation in space and time so the signal arrives at the
precise site r s as u s (t ) .

Let’s carry out these procedures with the Green’s function formalism.
108
Vol. 2 No. 2 August 2004
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

3. GREEN’S FUNCTION FOR THE DISCRETE CASE

Suppose we want to relate the values of u ( r , t ) at different places and times r, t and r ' , t ' . Because of the linearity
of the wave equation (1) and remembering that u ( r j , t ) = u j (t ) , it is possible to write [2] for the signal measured at
the site r j :
∞
u j (t ) = u j (t ) + ∑ ∫ G (o ) (r j , t ; r k , t ' ) Ak uk (t ' )dt ' ,
(o)
(2)
k ≠ j −∞

(o)
where G (r j , t ; r k , t ' ) is the free Green’s function and the Ak are complex scattering coefficients that contain the
full nonlinear interaction. The signal uk (t ' ) can be written in terms of the non-vanishing function S k (t ' ) defined by

⎧0 for t '∈ (−∞,0) ∪ (Τ, ∞)⎫

uk (t ' ) = ⎨ ⎬ (3)
⎩ S k (t ' ) for t '∈ [0, Τ] ⎭

Then, equation (2) can be written

Τ
S j (t ) = S (t ) + ∑ ∫ G ( o ) (r j , t ; r k , t ' ) Ak S k (t ' )dt '
(o )
j (4)
k≠ j o

But we can express the free Green’s function in terms of its Fourier Transform corresponding to the associated
frequency ω

∞
1
G (o) (r j , t; r k , t ' ) = ∫ Gω (r j , r k )eiω ( t −t ') dω
(o)
(5)
2π −∞

Thus equation (4) becomes

Τ ∞
1
S k (t ) = S k(o ) (t ) +
2π
∑
k≠ j
× ∫ ∫ Gω( o ) (r j , r k )eiω ( t − t ') dωAk S k (t ' )dt ' (6)
o −∞

This can be written as

∞ Τ
1
S k (t ) = S k( o ) (t ) +
2π
∑ Ak ∫ Gω(o) (r j , r k ) × [∫ e iω (t −t ') S k (t ' )dt ' ]dω
k≠ j
(7)
∞ o

Or also
∞
1
S j (t ) = S (jo ) (t ) +
2π
∑ Ak ∫ Gω(o) (r j , r k ) g k (ω )dω
k≠ j
, (8)
−∞

109
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

where
Τ
g k (ω ) = ∫ eiω ( t − t ') S k (t ' )dt ' (9)
o
That is, g k (ω ) is the Fourier transform of S k (t ) .
Also we have
∞
1
∫e
i ωt
S j (t ) = g j (ω )dω (10)
2π −∞

So equation (8) is now

∞ ∞ ∞
1 1 1
2π ∫ e g j (ω )dω =
i ωt

2π ∫ e g j (ω )dω +
iωt (o )

2π
∑ Ak ∫ Gω(o) (r j , r k ) g k (ω )dω
k≠ j
(11)
−∞ −∞ −∞

From this equation we obtain

g j (ω ) = g (jo ) (ω ) + ∑ Ak Gω(o ) (r j , r k ) g k (ω ) (12)

k≠ j
Or in vector form

g (ω ) = g (ω ) − G (ω ) g (ω )
(o) (o)
(13)
By factorizing,

g (ω ) = [1 − G (ω )]g (ω )
(o) (o)
, (14)
where

⎧ 0 if j = k ⎫
G j , k (ω ) = ⎨
(o)
⎬ (15)
⎩ Ak Gω (r j , r k ) if j ≠ k ⎭
(o)

Formally, we can invert equation (14)

g (ω ) = [1 − G (ω )]−1 g (ω )
(o) (o)
(16)
That is

g (ω ) = g (ω ) + G (ω ) g (ω ) + [G (ω )]2 g (ω ) + [G (ω )]3 g (ω ) + L
(o) (o) (o) (o) (o ) (o) (o)
(17)

The k component of equation (17) is

g k (ω ) = g k( o ) (ω ) + ∑ Ak Gω( o ) (r k , r t ) g t( o ) (ω ) + ∑∑ Al Gω( o ) (r k , r l ) At Gω( o ) (r l , r t ) g t(o ) (ω ) + L (18)

t t l

110
Vol. 2 No. 2 August 2004
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

Now, by substituting g k (ω ) from (18) into (12)

g j (ω ) = g (jo ) (ω ) + ∑ Ak Gω(o ) (r j , r k ){g k(o ) (ω ) + ∑ At Gω(o ) (r k , r t ) g t(o ) (ω ) + ∑∑ Al Gω(o ) (r k , r l ) At Gω(o ) (r l , r t ) g t(o ) (ω ) + L} (19)
k t t l

The brackets can be eliminated to give

g j (ω ) = g (jo ) (ω ) + ∑ Ak Gω( o ) (r j , r k ) g k(o ) (ω ) + ∑∑ Ak Gω(o ) (r j , r k ) At Gω( o ) (r k , r t ) g t( o ) (ω ) + L (20)

k k t

From this equation we obtain the discrete Neumann [3] series for the Fourier transform of the solution to the integral
equation (2)

g j (ω ) = g (jo ) (ω ) + ∑ At g t( o ) (ω ){Gω( o ) ( r j , r t ) + ∑ Gω(o ) (r j , r k ) Ak Gω(o ) (r k , r t ) + ∑ Gω(o ) ( r j , r k ) Ak Gω(o ) ( r k , r l ) Al Gω(o ) (r l , r t ) + L} (21)

t k k ,l

If the expression between brackets in (21) converges, it must be equal to the Fourier transform of the complete
Green’s function Gω (r j , r t ) , so that we can write

g j (ω ) = g (jo ) (ω ) + ∑ Ak Gω (r j , r k ) g k( o ) (ω ) (22)
k

Equation (22) can be written in vector form as

g (ω ) = [1 + G (ω )]g (ω )
(o)
, (23)
where
⎧ 0 if j = k ⎫
G (ω )i , j = ⎨ ⎬ (24)
⎩ A j Gω (r i , r j ) if j ≠ k ⎭

Equations (23) and (24) constitute our basic tools to perform the time inversion, that is, we can use equation (23) to
obtain experimental data of the components Gω ( r j , r k ) , because we know the Fourier transforms of the original
signals g k (ω ) , and we can measure the arriving signals g j (ω ) .
(o)

4. THE TIME INVERSION OF A SIGNAL

Let us carry out a time inversion. Suppose we have transmitted a signal burst from the site r s and that we have
recorded the arrived signals during a time Τ by the set {r b } . We can reverse the signal from each of the {r b } ; for
example, the signal u j (t ) recorded by r j can be emitted in reverse order to obtain a contribution at the original 1

place as
∞
us( o ) (t ) = ∫ G * (r j , T − t ' ; r s , t )u j (T − t ' )dt ' (25)
−∞
Or in terms of S j (t )

Τ
S (j o ) (t ) = ∫ G * (r j , T − t ' ; r s , t ) S j (T − t ' )dt ' (26)
o

111
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

We can express (26) in terms of the Fourier transform Gω (r j , r s ) :

*

∞
1
G * (r j , T − t ' ; r s , t ) = ∫ Gω (r
*
j , r s )e −iω (T −t '− t ) dω (27)
2π −∞

That is
Τ ∞
1
S s( o ) (t ) = ∫ ∫ Gω (r
*
j , r s ) × e − iω ( Τ − t ' − t ) dωS j (Τ − t ' )dt ' , (28)
o
2π −∞

which can be rearranged as

∞ Τ
1 *
∫− ∞ 2π Gω (r j , r s )e × [ ∫ eiωt ' S j (Τ − t ' )dt ' ]dω
− iω ( Τ − t )
S s( o ) (t ) = (29)
o

Based on properties of the Fourier transform, equation (29) is equivalent to

∞ Τ
1 *
∫− ∞ 2π Gω (r j , r s )e × eiωΤ [ ∫ e − iωυ S j (υ )dυ ]dω
− iω ( Τ − t )
S s( o ) (t ) = (30)
o

Or
∞
1
S s( o ) (t ) = ∫ 2π Gω (r , r s )eiωt g j (ω )dω (31)
*
j
−∞

(o)
And if we also put the Fourier development of S s (t ) :
∞ ∞
1 1
∫ e g s (ω )dω = ∫ 2π Gω (r
iωt (o ) *
j , r s )eiωt g j (ω )dω (32)
2π −∞ −∞

We get

g s( o ) (ω ) = Gω* (r j , r s ) g j (ω ) , (33)

where g j (ω ) is the Fourier transform of the signal received at r j , while g s (ω ) is the Fourier transform of the
(o)

signal that returned to r s after it has been reflected.

If we want the total contributions at r s from the set of transducers {r j } we must do the sum

Τ
S (t ) = ∑ ∫ G * (r j , Τ − t ' ; r s , t ) S j (Τ − t ' )dt ' ,
(o)
s (34)
j o

112
Vol. 2 No. 2 August 2004
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

which gives

g s( o ) (ω ) = ∑ Gω* (r j , r s ) g j (ω ) (35)
j

5. SENDING A MESSAGE TOWARDS A SITE.

Suppose now that our objective is to send a message from r j , that arrives at r s . Then we write [4]
∞
us( o ) (t ) = ∫ G * (r j , Τ − t ' ; r s , t )u j (t ' )dt ' (36)
−∞
From this and equation (27), we have

∞ Τ
1 *
∫−∞ 2π Gω (r j , r s )e × [ ∫ eiωt ' S j (t ' )dt ' ]dω
− iω ( Τ − t )
S s( o ) (t ) = (37)
o

This can be rewritten as

∞ Τ
1 *
S (t ) = ∫
(o)
Gω (r j , r s )e − iω ( Τ − t ) × [ ∫ e − iωt ' S *j (t ' )dt ' ]* dω (38)
2π
s
−∞ o

(o)
And by expressing S⋅ s (t ) and S j (t ' ) in terms of their Fourier series development we obtain

∞ ∞
1 e − iωΤ *
∫ e g s (ω )dω = ∫− ∞ 2π Gω (r j , r s )e g j (ω )dω
iωt (o ) i ωt *
(39)
2π −∞

This gives

g s( o ) (ω ) = e − iωΤGω* (r j , r s ) g *j (ω ) , (40)

where g j (ω ) is the complex conjugate of the Fourier transform of the message sent from r j , while g s (ω ) is the
* (o)

Fourier transform of the signal that was selected to arrive at r s .

For a batch of different messages from r j that go to the precise sites r s1 , r s 2 , r s3 , …, r s n , we have the sum

∑g q
(o)
sq (ω ) = ∑ e − iωΤGω* (r j , r s q ) g *j → q (ω ) ,
q
(41)

where j → q means that the signal begins at r j and its destination is r s q .

6. A SURVEY OF APPLICATIONS OF TIME INVERSION

As we said in the introduction, we developed a tool that can be applied to a broad class of phenomena where time
reversal is involved, and we gave the mathematical support. Let us recall several specific applications such as those
mentioned at the introduction. For the development of a device for destroying brain tumors or kidney stones [5] the
first step would be the localization of the abnormal sites with the help of an explorer emission and by means of the
above described inversion techniques which show and record the interesting points using a set of transducers. Then

113
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

the recorded signal is greatly amplified, sent, and returned backward in time toward the tumor or the stone to destroy
it. To improve the results we can repeat the exploration procedure several times before the final amplification.
Regarding the improvement of sonar underwater [5] detection, the procedure would basically consist again in the
emission of an exploration signal that is received and recorded tens of kilometers away from an initial sound source
by a series of transducers known generally as a Time Reverse Mirror (TRM); once the signal has been returned by the
TRM, another collection of transducers near the initial signal source record the remaining returned signal, resulting in
a very localized area in a very localized interval of time, and thus the TRM has been calibrated by the original signal
and any new obstacle can be detected easily. Another objective is to learn how to implement the process of sending
a message from one location to another very specific location. So, our immediate work is to implement an
experimental measure of the Green’s functions for simple geometries and simple arrangements of transducers. For
instance we can seek to improve the focus of a TRM reflected signal by increasing both the number of obstacles in
the signal trajectories and the time Τ during which the incident signal is recorded, phenomena that are explained by
an increment of the cross section of the receptors as the signal collides with each obstacle, thus improving the signal
definition, as was reported by other authors [4].

6. REFERENCES

[1] Colton, D. and Kress, R. Inverse Acoustic and Electromagnetic Scattering Theory, Applied Mathematical
Sciences 93. Springer-Verlag, Berlin, Heidelberg 1992, 1998.
[2] Sneider, R.K. and Scales, J.A. Time-reversed imaging as diagnostic of wave and particle chaos., Physical
Review E, 58, (5), 5668-5675,November (1998).
[3] Schiff, L.I. Quantum Mechanics. Mc Graw Hill International Editions. Auckland, Bogotá, Guatemala, Hamburg,
Lisboa, London, Madrid, México, New Delhi, Panamá, Paris, San Juan, Sao Paulo, Singapore, Sydney, Tokyo,
1968.
[4] Fink, M. Time Reversed Acoustics. Physics Today, 8, (3), 34-40, March (1997).
[5] Fink, M. Time Reversed Acoustics. Scientific American pages 67-73, November (1999).

Authors Biography

Juan Manuel Velázquez Arcos

Was educated at the Facultad de Ciencias of the Universidad Nacional Autónoma de México (FC-UNAM), where he
received his B.Sc., his M.Sc. in 1985 and his Ph.D. in physics in 1987. Currently he is Head of the Department of Basic
Sciences at the Universidad Autónoma Metropolitana - Azcapotzalco (UAM-A). Dr. Velázquez has lectured at FC-
UNAM, Escuela Nacional de Estudios Profesionales - Acatlán, Universidad Anáhuac, Tecnológico de Estudios
Superiores de Ecatepec and UAM-A. He is a full time professor of physics at UAM-A, and a member of the Area of
Physics of Irreversible Processes. He has been a member of two committees for the appraisal of academic work and
resources in physics at UAM-A. He has published 15 research papers and attended 25 physics meetings and symposia.
Dr. Velázquez is a member of the Sociedad Mexicana de Física and the American Association for the Advancement of
Science. His area of principal interest is mathematical physics, as applied to nuclear physics, acoustics, operator
algebra, Green functions, special functions, etc. His work has been awarded special mention by SOMI.

114
Vol. 2 No. 2 August 2004
 Time reversibility in acoustic signals, J.M. Velázquez, A. Ramírez & C.A. Vargas., 107-115

Alejandro Ramírez Rojas

Obtained his B.Sc. in physics at the Facultad de Ciencias at the Universidad Nacional Autónoma de México (UNAM).
He received his M.Sc. at the Escuela Superior de Física y Matemáticas of the Instituto Politécnico Nacional (ESFM-IPN),
and his Ph.D in physics at the Facultad de Ciencias of the Universidad Autónoma del Estado de México (UAEMex), with
specialty in physics of nonlinear phenomena. He has served since 1985 at the Universidad Autónoma Metropolitana -
Azcapotzalco (UAM-A), where he is currently a full time professor of physics at the Department of Basic Sciences, Area
of Physics of Irreversible Processes. He has lectured at the Facultad de Ciencias of UNAM, at the Tecnológico de
Estudios Superiores de Ecatepec, and also in the Master of Science Program in Computational Sciences of the
Fundación Arturo Rosenbleuth. He has served as coordinator of the core curriculum for engineering students, and as
an advisor in several divisional academic committees at UAM-A. Dr. Ramírez has participated in several national and
international physics congresses and seminars. Currently he focuses his research on the analysis of time series of
complex systems, as well as on the study of chaotic mappings. He has published several papers mainly on electrical
signals associated with earthquakes, and the application of deterministic chaotic and aleatory models to
electroseismical signals. He has been a member of the Sociedad Mexicana de Física for a decade.

Carlos Alejandro Vargas

Was born in Mexico City. He received his B.Sc. in physics at the Facultad de Ciencias of the Universidad Nacional
Autónoma de México (FC-UNAM), and his M.Sc. in physics at the Escuela Superior de Física y Matemáticas of the
Instituto Politécnico Nacional (ESFM-IPN), where he is currently working on his Ph.D. He is a member of the Mexican
National System of Researchers, where he has attained Level I status. In 1980 he served as Academic Technician at the
Instituto de Física of the Universidad Nacional Autónoma de México; from 1983 to 1991 he was an associate professor
of physics at the FC-UNAM, and since 1981 to this day he has been a lecturer at the ESFM-IPN. He was Head of the
Modern Physics Laboratory at FC-UNAM since 1987 to 1990. During 1997-1998 Professor Vargas was Head of the
Physics Area at the Universidad Autónoma Metropolitana - Azcapotzalco (UAM-A), where he is a full time professor of
physics. He is a co-founder of the Laboratory of Dynamical Systems (1991) and a founder and currently Head of the
Laboratory of Critical Phenomena and Complex Fluids (1998), both at UAM-A. He has directed several engineering
thesis and has published research papers on physics and physics education. At UAM-A he lectures mainly on optical
engineering, transducers and modern physics. His fields of interest include the physics of complex systems -specially
the study of mechanical, electrical and magnetohydrodynamic instabilities-, and theoretical work on electrodynamics
and quantum mechanics. He is a member of the American Association of Physics Teachers, the American Physical
Society, and the Sociedad Mexicana de Física.

115